Increasing prevalence of self-harm, suicidal behavior and suicidal ideation throughout a three-year period in the context of COVID-19 pandemic in Spain

had a great impact on mental health both in the general population and in individuals with preexisting mental disorders. Lockdown, social restrictions, changes in daily habits and limited access to health services led to changes in consultations in mental health services. This study aimed to determine changing trends in psychiatric admissions by the inclusion of adult patients admitted to the Emergency Department (ED) of Hospital Clínic of Barcelona between 2019 and 2021. Acute admissions, social issues and psychiatric diagnoses were compared between years, seasons and considering the interaction between both years and seasons. A total of 13,677 individuals were included in the analysis. An overall reduction in consultations to the ED and a higher proportion of acute admissions was observed in 2020 in context of the COVID-19 outbreak. Increased prevalence of sleeping disorders and substance use disorders was found in 2020. Self-harming behavior, suicidal thoughts and suicidal behavior showed an increasing tendency over time, with their highest rates in 2021. Prevention and management strategies should be considered in order to address increasing needs in mental health care.


Introduction
The coronavirus disease 2019 (COVID-19) pandemic has impacted different aspects of everyday life severely, causing important health and social consequences, including disease contraction, loss of employment, educational opportunities, freedoms, social contacts and supports, as well as loss of loved ones in many cases (Janoczkin et al., 2021).
The COVID-19 outbreak was acknowledged by the World Health Organization (WHO) as a pandemic in March 2020 (Cucinotta and Vanelli, 2020).This entailed the adoption of social restriction and quarantine in several countries in order to prevent the spread of the infection.In Spain, the first state of emergency and national lockdown was declared on March 14th, 2020, 2 and finished on June 21st, 2020 (Real Decreto 463/2020, de 14 de Marzo, 2020).
Lockdown and its aftermath not only led to important changes in daily habits, social contact and working and family routines, including loss of positive activities, but also to disruption of psychiatric counseling and treatment (Xiao, 2020), limited access to health services and resource scarcity (Torales et al., 2020).Some of these changes remained after the strict lockdown period or changed following the new COVID-19 waves (Janoczkin et al., 2021).At the present time, the majority of them have been progressively reestablished, whilst consequences in mental health seem to be still present (Monistrol-Mula et al., 2022).
Studies have demonstrated the impact of COVID-19 outbreak on mental health of the general population (Brooks et al., 2020;González-Sanguino et al., 2020;C. Wang et al., 2020), with perceived use of more adaptive coping strategies, such as following a routine or practicing physical exercise, than patients with previous mental illnesses (Solé et al., 2021), but also higher levels of depression, anxiety, distress and insomnia (Dragioti et al., 2022;Fico et al., n.d.;Wu et al., 2021;Yao et al., 2020).Also, changes in trends of psychiatric illnesses were observed, with recurrences or aggravation in patients with pre-existing mental disorders (Dubey et al., 2020;Iasevoli et al., 2021;Sagué-Vilavella et al., 2021;Solé et al., 2021;Taquet et al., 2021;Q. Wang et al., 2021).Indeed, patients with previous mental illnesses reported significantly more anxiety and depressive symptoms (Wu et al., 2021).Also, increased rates of suicide, self-harm and suicidal ideation during 2020 have been described (Farooq et al., 2021).All these variations caused the disruption of psychiatric care during the pandemic (Janoczkin et al., 2021).
Regarding the emergency department (ED) setting, during the early stages after the COVID-19 outbreak, increased prevalence of major depressive disorder, bipolar disorder and substance use disorder have been reported (Janoczkin et al., 2021).
Evidence regarding psychiatric conditions during the COVID-19 aftermath is scarce.First reports from a study focused on trends in psychiatric ED visits in adolescents showed an important increase in the number of admissions in 2021, mainly related to eating disorders, selfharm and suicide attempts (Fortea et al., 2023).A recent study also on children and adolescents showed an increase in ED visits and hospitalizations for suicidality from 2016 to 2019, with a temporary decrease in 2020 and a subsequent increase in 2021.Seasonal peaks in April were also reported (Kim et al., 2023).Other evidence reports an increase of 69 % in the mean monthly count of ED visits from the pandemic period among youth, especially in girls, which remained significant after adjusting for seasonality (Beaudry et al., 2022).
Nevertheless, limited data is available comparing adult consultation trends or mental health demands to EDs between the initial COVID-19 period with its aftermath.The pandemic and related measures have shown to affect suicide attempts differently across population groups, with higher risk not only in youth patients, but also in adults over 55 years (Kastner et al., 2022).A study comparing hospital-treated suicide attempts during the COVID-19 outbreak in 2020 and 2009 reported similar trends in adult patients, with a significant increase in adolescents during lockdown (Fernandez-Fernandez et al., 2023).However, potential variations throughout the subsequent months were not assessed.
Lifestyle changes associated with the emergence of COVID-19 seemed to be associated with sleeping changes.Indeed, a crosssectional study reported that anxiety levels about the pandemic were significantly associated with insomnia severity (Sugawara et al., 2023).Literature on the progression of sleeping disorders during the pandemic and its aftermath is scarce.
With regard to substance use disorders (SUD), the extreme restrictions imposed during the pandemic seemed to have detrimental impacts on vulnerable individuals.A recent study reported that the number of subjects who used heroin, cocaine, and MDMA decreased during lockdown, increasing in the aftermath up to pre-COVID levels.Cannabis use was less impacted, remaining stable throughout the pandemic, and alcohol use increased significantly during the lockdown and remained at significantly increased levels (Gili et al., 2023).Given the psychiatric and medical comorbidities, as well as social problems associated, the impact of COVID-19 on SUD should be seriously considered.
Even though the effects of the pandemic in general populations' mental health have been described (Brooks et al., 2020;González-Sanguino et al., 2020;C. Wang et al., 2020), little is known regarding patients with previous severe mental illnesses.
Previous literature has also widely reported the impact of seasonality on mental health disorders.Specifically, studies focusing on ED visits have shown winter peaks for depressive symptoms, summer peaks for suicidal activity, and spring/summer peaks for manic-related hospital admissions.Suicidal behavior seems to increase in spring and summer compared to winter (Della et al., 2023;Geoffroy et al., 2014a).This suggests that environmental conditions may exert a fundamental role on psychopathological conditions affecting biological vulnerability (Aguglia et al., 2017).
The aim of this study is to report the consultation trends of adults to the Psychiatric ED to one of the main third-level hospitals of Barcelona, Spain, throughout a three-year period, including the year previous to the COVID-19 outbreak, the lockdown period and further pandemic periods, including recent times in which most social restrictions have been reverted.Specifically, we aimed to examine how consultations due to specific psychiatric diagnoses, including suicidal-related behaviors, SUD and severe mental illnesses, differ among seasons and years of consultation.This study also aims to determine the tendency of visits according to different diagnoses along the different periods.The main hypothesis of this study was that an increase in suicide-related behaviors might be observed during the initial COVID-19 periods, with a tendency to decrease in subsequent periods, and that patients with severe mental illnesses and SUD might be at risk for increased ED visits during the earliest pandemic periods.

Participants
The analyzed data corresponded to records from patients aged 18 years or older admitted to the Psychiatric Emergency Service of the Hospital Clínic of Barcelona.The records included all the patients attending this service between 2019 and 2021.Ethical approval was provided by the Hospital Clínic Research Ethics Committee (protocol HCB/2022/0927).
Patients from which no gender or sex was provided were excluded from the study (n = 23), since the analyses were adjusted by these variables.Considering that all the admissions were recorded, patients with more than one admission in this period were included all the times they attended.

Sociodemographic and clinical measures
Some of the variables collected for analysis included age, gender, acute admission, Spanish nationality, presence of social problems and/ or medical conditions, psychiatric diagnosis, year and meteorological season at consultation.

Diagnostic variables
Primary diagnostic at consultation was considered for analysis.Mood disorders included bipolar disorder, major depressive disorder, adjustment disorder or non-specified mood disorder.Other diagnoses considered were anxiety disorder, sleeping disorder, conduct disorder, intellectual disability, post-traumatic stress disorder, eating disorder, social problems or issues, self-harming behavior, suicidal ideation, established when ideas, or ruminations about the possibility of completing suicide were present (Hubers et al., 2018), and suicidal behavior, which included suicide attempt or intentional drug poisoning, not considering substance abuse.Psychotic disorders encompassed schizophrenia, schizoaffective disorder, delusional disorder and non-specified psychosis.Substance use disorders included use of alcohol, opioid, cannabis, benzodiazepines, cocaine, inhaled drugs and stimulants.All diagnoses were conducted by assistant psychiatrists and collected according to ICD-10 criteria into the informatic system before patients' discharge (World Health Organization (WHO), 1993).

Seasonal variables
For the whole sample, season and year at consultation was

Statistical analyses
In the descriptive analysis, quantitative variables were expressed by mean and standard deviation.Results from categorical variables were shown as frequencies (number and percentage).Sociodemographic and clinical variables were analyzed.T-test and Chi-square tests were used to study the differences between groups according to the year they attended the ED.To assess the different outcomes over time and seasons, we use generalized linear models with time x season interaction adjusted by sex and age.To compute each adjusted prevalence, we used the library emmeans (v.1.8.3).Secondary analyses were conducted to assess variations between three specific periods according to lockdown: before lockdown (from January 1st 2019 to March 14th 2020), the lockdown period (from March 14th 2020 to June 21st 2020) and the period after lockdown (from June 22nd 2020 to December 31st 2021).The effects of lockdown were assessed with logistic regression analysis according to the three main periods using the turkey method for multiple comparisons.All statistical analyses were performed by the use of a confidence interval of 95 %, considering significative differences when p < 0.05.Analyses were conducted with SPSS version 25.0 and R version 4.1.2(R Core Team, 2020).

Characteristics of the sample
A total of 13,677 admissions to the psychiatric emergency service of Hospital Clínic of Barcelona between 2019 and 2021 were included in the analysis.Socio-demographic and clinical characteristics of the whole sample are available in Table 1.
Among all patients included for analysis, 4814 were visited in 2019, 4007 in 2020 and 4856 in 2021.From the total sample, 49.9 % were female, with a significantly different proportion between years, being the lowest in 2020 (47.7 %) and highest in 2019 and 2021 (50.8 % and 50.7 % respectively).Mean age of the sample was 40.47 years (SD 15.83), with no significant differences between years.
The percentage of Spanish individuals varied significantly between years, with a higher proportion in 2019 (79.8 % in 2019 versus 75.4 % in both 2020 and 2021).Also, the proportion of patients with medical comorbidities varied, with the highest frequency in 2019 (3.6 %) and the lowest in 2021 (2.7 %).
When the proportion of patients visited at the ED according to each meteorological season was compared between years, significant differences were found except for summer, with a proportion of consultations significantly lower in spring 2020 and higher in winter 2020 and autumn 2021.Main effects of season, year and interaction of both, adjusted by sex and age, are represented in Table 2.

Acute hospital admission
Significant differences were observed in the proportion of acute hospital admissions in terms of year of consultation (p < 0.001) (Fig. 1A).Specifically, a higher proportion of patients required acute hospital admission after attending the ED in 2020 (27.2 %) compared to 2019 and 2021 (24.1 % and 23.0 % respectively).The proportion of acute hospital admissions also showed significant differences according to the season (p < 0.001).Lastly, it showed significant differences when the interaction between year and season were considered (p < 0.001).Particularly, in the spring of 2020, a significant increase in the proportion of acute admissions (34.0 %) among total consultations was observed.

Psychotic disorders
Consultations of patients with psychotic disorders were significantly higher in number and proportion in 2019 (20.6 %) than 2020 or 2021 (18.3 % in both; p < 0.01), with a specific increase in spring 2019 (22.2 %).No differences were found when schizophrenia, schizoaffective disorder or delusional disorder were studied separately.In contrast, when only the diagnosis of schizophrenia was considered, no differences were shown between years, but our sample showed seasonality (p < 0.05) with an increase in consultation rates in winter and a reduction in summer.This seasonal pattern was not observed in schizoaffective disorder or in delusional disorder, although in the last one, an increase in the proportion of visits was observed in autumn of 2020 compared to all other periods.

Mood and anxiety disorders
The frequency of mood disorders varied significantly among the three years (p < 0.001), with a proportion significantly higher in 2019 (17.4 %) compared to 2020 and 2021 (14.6 % and 15.3 % respectively), and when the combination effect between years and seasons was assessed (p < 0.01), but not when seasons were compared (p = 0.06), having spring and summer of 2019 the highest proportions (18.3 % and 18.6 % respectively).

Bipolar disorder
Overall consultations of patients with bipolar disorder did not vary significantly among years or seasons.
Specifically, no differences between years were observed in manic episodes, but a seasonal pattern was shown in our sample, with the lowest proportions in winter of the three years (p < 0.05).
Regarding bipolar depression, no significant differences among years or seasons were observed.

Major depressive disorder
The proportion of patients attending the ED with major depressive disorder varied among years (4.7 % in 2019 versus 3.3 % in both 2020 and 2021; p < 0.001) and when both years and seasons were assessed (p < 0.05).Specifically, 2019 was the year with the highest proportion and summer the season with highest frequency in this year (5.04 %).Nevertheless, no significant differences between seasons were observed.

Anxiety disorders
The proportion of patients consulting with anxiety disorders was significantly lower (19.8 %) in 2020 compared to 2019 and 2021 (23.3 % and 23.0 % respectively), with no differences between different seasons.

Substance use disorders
The proportion of visits related to substance use disorders varied among seasons (p < 0.05) and years (p < 0.001), and also when the interaction of both factors was assessed (p < 0.05) (Fig. 1B).Specifically, the rate of patients consulting because of substance use disorders was higher in 2020 (13.5 %) compared to the other years (10.8 % in 2019 and 11.1 % in 2021), but the total number of patients visited did not vary.
Summer and winter of 2020 were the seasons with significantly highest proportions of patients with substance use disorders.
When this subsample was split into groups according to the drug used, a proportion significantly higher was found in patients with cannabis use disorder in 2020 (3.2 %) compared to 2019 and 2021 (2.6 % and 2.1 % respectively; p < 0.01).In addition, significant differences were found between seasons (p < 0.05) and the interaction of both (p < 0.05).A significant increase in the proportion of consultations was observed in summer of 2020 (3.0 %), being this year the one with the greatest prevalence of this diagnosis.In 2021, a tendency towards a smaller proportion was observed, with the lowest rate in winter (0.8 %).
The proportion of patients visited with alcohol use disorder did not vary significantly among years or seasons.
Stimulant use disorders had also variations between years (p < 0.001), with a significantly lower proportion found in 2019 (1.5 %) compared to 2020 (2.5 %) and 2021 (2.9 %).Thus, an increasing tendency over the three years was observed, with no seasonal patterns observed.

Suicidal thoughts
Emergency visits regarding suicidal thoughts showed significant differences between years (p < 0.001) and between both years and seasons (p < 0.05), but not when seasons were compared (p = 0.31).Our data showed a significant increased prevalence in 2021, specifically in spring and summer (3.2 % and 4.2 % respectively), compared to the other years (Fig. 1C).

Suicide attempt or poisoning and self-harming behaviors
Even though seasonality was not a differential factor for suicide attempts or self-poisonings (p = 0.17), significant differences were found between years (p < 0.01).Concretely, spring and autumn of 2020 were the periods with highest proportions (2.92 % and 3.82 % respectively),  and winter of 2019 the season with the lowest (1.36 %) (Fig. 1d).
Even though no differences between seasons were observed in the proportion of self-harming behaviors, 2020 was the year in which highest proportions were observed, concretely in autumn (Fig. 1e).Significant differences were found between years (p < 0.01), with an increase in 2021, but not in the interaction between years and seasons (p = 0.013).

Other conditions
No significant differences were observed in the proportion of  consultations of patients with borderline personality disorder, posttraumatic stress disorder or eating disorder.In contrast, significantly different proportions were found in sleeping disorders, conduct disorder and intellectual disability.Among these, sleeping disorders and conduct disorders displayed higher proportions in 2020, and intellectual disability in 2021, compared with the other two years.
With regard to the presence of social problems, no significant differences were observed between the three years.However, it followed a seasonal pattern, with an increase of these consultations in winter (p < 0.01), being differences in the interaction of years and seasons also significant (p < 0.05).

Influence of lockdown
Differences between three different periods considering lockdown were studied.The adjusted prevalence for the main conditions and pairwise comparisons between all periods are represented in Table 3. Significant differences were found in the prevalence of acute admissions when the lockdown period was compared with the periods before and after lockdown (p < 0.001 for both comparisons), with a higher prevalence of admissions during lockdown (33.7 %) compared to the rest of periods (24.1 % before lockdown and 23.9 % after lockdown).Patients with mood disorders were less likely to attend the ED (11.7 %) compared to the periods before (16.9 %, p < 0.001) and after lockdown (15.5 %, p = 0.010).The same was observed with anxiety disorders, with 15.9 % of prevalence during lockdown and 23.1 % before lockdown (p < 0.001) and 22.2 % after lockdown (p < 0.001).Substance use disorders were more prevalent in the ED during lockdown compared with the previous period (p = 0.010) and also between the periods and after lockdown (p = 0.038).Suicidal thoughts-related visits experimented an increase in prevalence after lockdown compared to the period before lockdown (p < 0.001); the same was observed with self-harming behaviors (p < 0.01).

Discussion
The present investigation shows the impact of the COVID-19 pandemic in the psychiatric ED in early and also later stages through a retrospective cross-sectional study, with variations in trends throughout the three years studied.This study found a reduction in the total number of ED visits in 2020, being in 2021 similar to 2019.
In 2020, the year in which COVID-19 outbreak was acknowledged as a pandemic and restrictions were implemented, even though the total number of acute hospital admissions did not increase, the proportion was 27.2 % of the total patients attending the ED, which was significantly higher compared to the previous and following years.This could mean that visits from the most severe patients might have been stable despite social restrictions.
Reasons for consultations also varied significantly between the different periods.In contrast to what has been described in the literature regarding outpatients (Solé et al., 2021;Wu et al., 2021), the proportion of mood disorders attending our psychiatric ED was higher in the prepandemic period compared to the other periods.In addition, the proportion of patients consulting with anxiety disorders was lower in 2020.Interestingly, our results showed a lower prevalence of mood and anxiety disorders in the ED during the lockdown period.
The disparity of these results with previous literature not focused on psychiatric ED might be due to reduced help-seeking behaviors during the first periods of the pandemic in patients suffering anxiety or mood disorders secondarily to quarantine measures and mobility restrictions.
The lack of significant differences in the proportion of eating disorders between the three years contrasts with recent data from adolescent samples (Fortea et al., 2023).Also, suicidal behavior was found in higher frequency in 2019, whereas suicidal ideation was more commonly seen in the ED in 2021.Our results also demonstrated a significant increase in the prevalence of consultations related to suicidal thoughts and self-harming behavior after lockdown compared to the period before lockdown.In the same geographical area (Catalonia), an overall decrease of suicide-related thoughts and behavior was seen in community-based health care centers, along with an increase of completed suicides (Pérez et al., 2022).Hence, it appears that the pandemic lowered the capacity of community-based care to detect suicidal behavior, causing more visits to the ED and more suicides.As described in previous studies (Dragioti et al., 2022;Wu et al., 2021;Yao et al., 2020), the proportion of consultations regarding sleeping disorders and also conduct disorders was higher in 2020, which could have been related with the drastic change in daily routines and reduced in-person social contact (Monteleone, 2021).
According to our data, substance use disorders seemed to maintain similar number of visits to the ED, with increased proportion in 2020 given the lower number of total visits.Particularly, cannabis use disorder showed increased frequency in 2020, which might be displaying increased use during the lockdown and its aftermath, showing a reduction later in 2021.
No significant differences were found when bipolar disorder, schizophrenia, schizoaffective disorder or delusional disorder were compared in frequency of visits between years, which suggests a tendency of severe mental disorders to maintain their prevalence, even though the increased rate in admissions in 2020 could be explained by worse psychopathological status in this group of patients at consultation (Vita and Barlati, 2022).
The seasonal pattern observed in the percentage of acute admissions, with a proportion as high as 34.0 % during 2020 spring, also highlights the impact COVID-19 outbreak had on mental health care.
Regarding severe mental illnesses, bipolar disorder presented a seasonal pattern, which is independent of COVID-19, in line with previous evidence (Fico et al., 2021;Geoffroy et al., 2014b), with lowest proportion of manic episodes in winter.Also, schizophrenia rates increased in winter and decreased in summer, with a milder effect.Increased proportion of social issues in winter might be explained in some cases by the unfavorable weather and homelessness.
An increased proportion of consultations related with substance use disorders was observed in 2020 summer, especially related with cannabis use, coinciding with the pandemic period just after lockdown.This might be related directly with reduced social contacts and home isolation, with permission at the same time for seeking substances outside.
Some studies have reported increased suicide attempts and selfharming behaviors during the COVID-19 pandemic (Farooq et al., 2021).From our sample, concerning suicidal thoughts, a tendency to increase throughout the three years was observed, not only after the COVID-19 outbreak, but also during its aftermath.With respect to suicide attempts or self-poisonings, 2019 showed significantly lower proportions compared to 2020 and 2021.The increase observed in autumn 2020 seems to compensate the reduction in summer 2020, coinciding with one of the earliest pandemic periods.The same occurred with self-harming behaviors, with progressive increase over the different years, a reduction in 2020 summer and an exceptional increase in 2020 summer, maintaining higher prevalence in 2021 compared to the previous years.The increasing tendency of suicidal ideation, suicide attempts and self-harming behavior observed in 2021 might be consequences of the pandemic itself but also the persistence of some restrictions, limited social interactions, changes in daily routines and difficulties in the access to mental health services in the community, together with less fear of the virus.
Clinicians should pay special attention to the recent tendencies in order to face potentially increased mental health demands and provide effective specialized care, and also develop strategies focused on suicide prevention.Future studies will elucidate long-term consequences of the pandemic, especially in anxiety and depressive disorders, self-harm, suicidal ideation and suicide attempts (Pompili, 2021).
Among the strengths of this study, a large sample of adults visited in a third-level hospital ED was included, and a considerable period of time was assessed.All patients included in the study were evaluated by a trained psychiatrist.In addition, this study controlled for potential confounders, such as sex, age, seasons and years.
However, our study comes with some limitations.First, the crosssectional design of the study cannot reveal causal dynamics between considered variables.Also, it is a naturalistic and pragmatic study in which only the main diagnosis was provided, even though a number of patients might have comorbid conditions.Robust statistical analyses of each diagnosis were performed comparing proportions between periods and not with total number of patients, and also not considering the different COVID-19 waves.Intentionality of poisonings was not differentiated in this sample.Lastly, since our findings provide from a single center, they may not be generalizable to other regions in Spain or worldwide, especially those in which restriction measures or COVID-19 incidence were different.Longitudinal studies including longer periods of time after the COVID-19 outbreak will increase clinicians' knowledge regarding the tendency of the most common reasons for consultation in Emergency Services.
In conclusion, the COVID-19 pandemic led to a significant reduction in the overall consultations in 2020, with a higher proportion of acute admissions and similar proportions of consultations of patients with severe mental disorder.A significant increase in the proportion of patients with substance use disorder, specifically cannabis, was observed in summer of 2020, corresponding to the period after lockdown with persistence of social restrictions.The increasing tendency of suicidal ideation, suicide attempts and self-harming behavior observed over time might have been conditioned by health and mental consequences of the pandemic, social isolation and lifestyle changes, as well as the reduced accessibility to health care.Future studies will elucidate long-term consequences of the pandemic, especially in anxiety and depressive disorders, self-harm, suicidal ideation and attempts.Clinicians should be prepared to face increased mental health burden and design effective preventive and therapeutic strategies.

Fig. 1a .
Fig. 1a.Estimated prevalence of acute hospital admission adjusted by age and sex.Fig 1b Estimated prevalence of substance use disorder adjusted by age and sex.Fig 1c.Estimated prevalence of suicidal ideation adjusted by age and sex.Fig 1d Estimated prevalence of poisoning or suicide attempt adjusted by age and sex Fig 1e.Estimated prevalence of self-harming behavior adjusted by age and sex.

Table 1
Socio-demographic and clinical characteristics of the sample according to the year attending the Emergency Service.

Table 2
Main effects of year, season and interaction of year and season on different diagnoses.

Table 3
Adjusted prevalence for period (before lockdown, lockdown, after lockdown) and pairwise comparisons between periods.